Soil erosion is a complex process involving material detachment, transportation (mainly by water, occasionally by wind), and eventual deposition when energy wanes. Human activities like tillage and construction can exacerbate soil erosion. Various forms of water erosion, such as sheet erosion, pedestal erosion, rills, piping, and gullying, are recognized. This pioneering study aims to comprehensively model water erosion across Taiwan at a 20 m spatial resolution, a departure from previous research focusing on smaller scales. Using the Revised Universal Soil Loss Equation (RUSLE) model, it seeks to examine the significant issue of soil erosion in Taiwan beyond agricultural areas and enable cross-regional comparisons. A large number of stations and the most recent data were used to establish the distributions of the rainfall runoff erosivity factor and the soil erodibility factor. In addition, we used the Google Earth Engine (GEE) to calculate the Normalized Difference Vegetation Index (NDVI) and a locally derived empirical equation to compute the cover-management factor. The topographic factor was determined using the System for Automated Geoscientific Analyses (SAGA). The support practice factor was analyzed using two different methods using datasets on World Urban Areas and Global Urban Boundaries from the literature. The analysis showed that despite the difference in the support practice factor, Taiwan’s resulting average yearly soil erosion rates are very similar (200.7 Mg ha−1 year−1 and 207.4 Mg ha−1 year−1). The amounts were validated against five watersheds that were instrumented with erosion pins. With prediction ratios ranging from 1.04 to 1.82 across four of the five watersheds, our findings provide empirical support for the alignment of our model with soil erosion pin measurements, especially within the Tsengwen reservoir watershed. However, it is noteworthy that these results also exhibit a tendency towards conservative estimations in the remaining watersheds. Our calculated estimate, falling within the range of 201–207 Mg ha−1 year−1, plausibly represents the upper limit of mean soil erosion in Taiwan. This assertion is predicated on the deliberate omission of local soil conservation measures from our study, a decision necessitated by the absence of comprehensive and detailed island-wide data. Despite this limitation, our results instill confidence in the robustness of our methodological approach, thereby suggesting that our estimation of soil erosion in Taiwan provides a reliable approximation.